v1.10.1

What's Changed

• ci: make the roundtrip fidelity benchmark a blocking gate by @thomas-villani in #165
• bench(startup): measure CI runner variance before gating on it by @thomas-villani in #166
• test(perf): ratchet what import all2md pulls in eagerly by @thomas-villani in #167
• ci: gate cold start against a committed baseline by @thomas-villani in #168
• perf(options): re-export lazily instead of importing every submodule by @thomas-villani in #169
• bench(spread): exclude ungated scenarios from the tolerance floor by @thomas-villani in #170
• ci(corpus): gate the reproducible half of the corpus on failure set by @thomas-villani in #171
• ci(corpus): record the corpus fidelity baseline by @thomas-villani in #172
• fix(tests): stop the benchmark conftest from hijacking -m by @thomas-villani in #173
• fix(corpus): stop caching a failed download as an empty result by @thomas-villani in #174
• test(perf): benchmark the formats, stop pretending to gate them by @thomas-villani in #175
• feat(action): ship the conversion-quality gate as a GitHub Action by @thomas-villani in #176
• fix(tests): use tmp_path_factory for TestSplitCLIE2E to avoid parallel races by @rkfshakti in #187
• chore(deps): Bump actions/setup-python from 6 to 7 by @dependabot[bot] in #188
• chore(deps): Bump actions/download-artifact from 7 to 8 by @dependabot[bot] in #189
• fix(markdown): require a delimiter row before parsing a table by @thomas-villani in #190
• fix(corpus): retry the two large downloads, and report formats that matched nothing by @thomas-villani in #191
• fix(ci): make the lint gate able to fail by @thomas-villani in #192
• docs(performance): source the numbers, or remove them by @thomas-villani in #193
• chore(release): prepare 1.10.1 by @thomas-villani in #194

New Contributors

• @rkfshakti made their first contribution in #187

Full Changelog: v1.10.0...v1.10.1